COMPARISON OF 2 BINDING-EQUATIONS FOR PREDICTION OF THE CONCENTRATIONOF UNBOUND VALPROIC ACID IN THE SERUM OF ADULT EPILEPTIC POLYTHERAPY PATIENTS

Because binding of valproic acid to plasma proteins affects the effica cy of the drug in the treatment of epilepsy (only the unbound fraction of the drug is effective) we have compared two methods which use diff erent binding parameters to predict the in-vivo concentration of unbou nd valproic acid in serum. The study was performed on 46 serum samples from 29 polytherapy adult patients with epilepsy. Mean prediction err or, mean absolute prediction error and root mean squared error were ca lculated for each method; these values served as a measure of predicti on bias and precision. The mean absolute prediction errors and root me an squared errors for the two methods were similar in magnitude (Metho d 1, mean absolute prediction error = 10 . 0 mu M, root mean squared e rror = 15 . 0 mu M; Method 2, mean absolute prediction error = 10 . 3 mu M, root mean squared error = 13 . 5 mu M). Method 2 had a general t endency to over-predict unbound valproic acid; both methods had a tend ency to over-prediction for total concentrations above 500 mu M. Metho d 1 had a tendency to under-prediction at total concentrations below 2 50 mu M. Within the total concentration range of valproic acid investi gated, Method 1 was superior to Method 2 for prediction of unbound ser um valproic acid. Our approach using Method 1 may be useful for predic tion of unbound serum valproic acid concentration in patients with tot al valproic acid concentrations ranging from 250 to 500 mu M; Method 2 may be useful for patients with total valproic acid below 500 mu M. O ur results suggest that there is wide and unpredictable variability in valproic acid binding to serum proteins among study populations.